考虑网评的电商产品组选择问题研究

发布时间：2018-05-17 19:50

本文选题：产品组选择 + 在线网评　；参考：《西南财经大学》2014年硕士论文

【摘要】：产品组选择问题是存在于零售行业、航空业、家用电子行业等行业内的基础性问题,也是主要的运作问题。产品组选择即：商家在资金、仓储、店面等资源有限的情况下从数量众多的产品中选择一组产品来进行销售的过程。目的旨在通过对这些限制条件的权衡达到产品销量的提升和利润的最大化。零售商希望在合适的时间,合适的地点,提供合适的商品组合来满足消费者的需求。首先,做到这点不容易,其次,真的做到了这一点也并不能保证商家的产品组选择一定正确,因为产品组选择是由每一个销售时点以及每一个零售店具体的销售情况决定的,所以零售商应该具体问题具体分析,不断调整产品策略来适应季节的变迁、新产品的引入、消费者喜好的改变等因素。所有因素里,无疑,经济环境的影响是深远的。在互联网经济不断发展的今天,越来越多的企业意识到网络信息传播的重要性,与网络密切打交道的电商企业更是不例外。作为最近几年蓬勃发展的一个新兴行业,电商企业已逐渐显示出它的优势和生命力。网上零售商的经营者享受着购物信息在消费者之间充分流转的同时,必须时刻警惕网络信息的负面影响,只有研究并关注网上消费者的消费行为,掌握他们的网评信息因素,才能化不足为长处,为自己的发展找到方向。同传统购物模式相比,网上消费者的购买决策会更多受到网络产品信息的影响,相对来说,他们会更加理性地比较各类网评信息,因为网络购买受外界因素影响较小,购买者可以面对电脑屏幕更全面地浏览商品的网评信息。网评信息主要地包括网评得分,网评有用性,得分方差,网评文字部分,网评总量。也有很多学者就这些指标对产品销量的影响展开研究,然而,本文发现,在关于网评维度的研究里,学者并没有就某一项对于商家销量的影响达成共识,这使得网评研究具有广阔发展空间的同时,选取合适的网评指标作为研究对象成为棘手之事。本文参照Sun(2010)和Nan Hu(2010)的研究观点,选取网评得分(Valence)和评分标准方差(Variance)这两项网评维度作为研究零售商产品组选择问题的影响因素。而企业的资源是有限的,经营者总是希望能够通过正确地配置资源,达到收益的最大化。不同的产品各有特点,如何利用好这些产品之间的组合选择,特别是加入了无形的网评信息,这使得产品评估信息更加多元化,对诸多网上零售企业来说,做到合理的权衡是一件难事。消费者对产品的决策是一个离散选择的过程,因此离散选择模型(Discrete Choice Model)在这类问题中得到广泛应用。本文选取最基础也是运用最广泛的MNL (Multinomial Logit)模型来模拟顾客的选择过程。长久以来,在零售商产品组合优化问题中,解空间是随着可选产品集合数量成指数增长的,这是一个NP-Hard司题,一般很难精确地求出最优解,因此寻求一个有效的近似解算法是十分必要的。实践证明,遗传算法正是启发式算法里应用比较广泛、求解比较得力的一种算法,但针对不同的优化问题,也会存在严重的局限性,尤其表现在产品组合优化的求解问题中。因此本文采用的是改进的基于贪心策略的遗传算法,以期望能够克服单纯遗传算法的某些弱点,来更快达到最优解。追溯相关产品组选择问题的研究,本文发现,研究主要集中在库存与产品组选择的关系,缺货替代与消费者需求之间的关系,产品组选择的最优算法或者近似最优算法等,没有学者将消费者对于产品的网评信息纳入到产品组选择问题中来。因此,在网络经济时代,笔者认为将二者结合起来,做出尝试性的探讨是有意义的。本文的创新之处主要体现在：1.在离散选择模型的基础上加入了除传统的价格影响因素外的网评得分与标准方差的新元素,以适用于网络环境的新背景,提出了具体的消费者需求评估模型,即改进版的MNL模型,拓宽了MNL模型的理论应用范围；2.创新性地将产品分类计划的研究问题考虑到时下热门的电子商务环境中来,试图发现消费者网评信息对于电商产品组选择策略的影响,希望能够做出一点有益的帮助。在导师的指导下,本人通过使用改进的MNL离散选择模型,将顾客效用表达式内涵地包括了产品价格、网评得分与评分标准方差,并使用基于贪心策略的遗产算法将零售商产品组选择问题转化为生物遗传语言,进行MATLAB编程,通过模型的构建,分析了不同参数设置下的结果变化情况,得出了如下结论： 1.产品的网评得分对于电商的产品组选择有着积极,正向的影响,顾客越在意产品的网评,也就越愿意选择网评得分高的产品,这样就会有更多的顾客将资金用在几种评分高的产品,这就导致了虽然选择的产品少了,却会有更大的利润。建议零售商在顾客逐渐重视网评得分的情况下,应该逐渐减少产品的种类,以便于顾客购买产品。 2.网评得分标准方差与产品组选择之间呈现出负向的关系,也就是说,随着消费者对方差影响敏感系数的增大,商家的利润减小,产品种类数增大。这说明如果顾客逐渐看重商品网评评分的集中程度,那么消费者选择商品的概率就减小。商家为了得到更多的利润,就不得不增加产品的种类。 3.关于算法实现,在本文算例中证明,遗传算法中种群大小的改变会或多或少影响最终商家选择结果的变化。在其逐渐变大的过程中,遗传进化代数总体变少,但每一次迭代的速度变慢了,收敛时间变长了,总体能在在较少时间内找到问题最优解。本文在网络时代消费者日益受到网评信息影响的背景下,构建了在线评论影响商家产品选择的模型,研究了在线评论对网络零售商产品决策的影响,建议了电商企业重视相关网评信息的收集与评估,丰富了PAP问题的研究范围,具有一定的创新性。但是由于时间和能力的制约,论文无法全面分析在线评论对商家决策的影响,只能从局部的几点出发,具有一定的片面性。尽管论文关于在线评论对电商企业产品组选择的影响的研究不够深入透彻,但毕竟是一次有益的尝试,希望能为以后的深入研究提供有效的参考,并具备一定的理论价值和实践意义。
[Abstract]:The choice of product group is the basic problem that exists in the retail industry, the aviation industry, the home electronics industry and so on. The product group chooses to choose a group of products from a large number of products in the case of limited resources such as funds, warehouses and stores. The trade-off between these constraints can achieve the promotion of product sales and the maximization of profit. The retailer wants to meet the needs of the consumer at the right time, the right place, and the suitable product mix. First, it is not easy to do this. Secondly, it does not guarantee the choice of the product group of the merchant to be correct. Because the selection of the product group is determined by each time point of sales and the specific sales situation of each retail store, the retailer should analyze the specific problems and adjust the product strategy to adapt to the changes of the season, the introduction of new products, the change of consumer preferences, and so on. All factors, undoubtedly, the impact of the economic environment is Profound.
With the continuous development of the Internet economy, more and more enterprises are aware of the importance of the communication of network information. The business enterprise which is closely related to the network is no exception. As a booming industry in recent years, e-commerce enterprises have gradually shown its advantages and vitality. The operators of online retailers enjoy it. While shopping information is fully transferred among consumers, we must always be vigilant on the negative effects of network information. Only studying and paying attention to consumer behavior on the Internet and mastering the information factors of their online reviews can make the lack of the strengths and find the direction for their own development.
Compared with the traditional shopping model, the purchase decision of online consumers will be more affected by the information of the network products. Relatively speaking, they will compare all kinds of online reviews more rationally, because the network purchase is less influenced by the external factors and the buyers can browse the online reviews more comprehensively on the computer screen. The key points include the score of net reviews, the usefulness of the net reviews, the variance of the score, the text of the net reviews, the total amount of the net reviews. There are also many scholars on the impact of these indicators on the sales of products. However, this paper finds that in the study of the dimensions of the net reviews, the scholars have not reached consensus on the impact of a particular item on the sales of the business, which makes the online review study At the same time, it is a thorny thing to choose the suitable index of net reviews as the research object. This paper, referring to the research views of Sun (2010) and Nan Hu (2010), selects the two dimensions of net evaluation score (Valence) and score standard variance (Variance) as the influencing factors of the study on the selection of the retailer's product group.
And the resources of the enterprise are limited, and the operators always hope to maximize the income by correctly configuring the resources. The different products have their own characteristics, how to make use of the combination selection between these products, especially the invisible net evaluation information, which makes the product evaluation information more diversified and to many online retail enterprises. It is difficult to make a reasonable tradeoff. The decision of the consumer to the product is a discrete selection process, so the discrete selection model (Discrete Choice Model) is widely used in this kind of problem. This paper selects the most basic and most widely used MNL (Multinomial Logit) model to simulate the process of customer selection. In the problem of retailer portfolio optimization, the solution space is increasing exponentially with the number of optional product sets. This is a NP-Hard division problem. It is generally difficult to find the optimal solution accurately. Therefore, it is necessary to seek an effective approximate solution algorithm. Practice has proved that the genetic algorithm is widely used in heuristic algorithms. In order to solve the problem of different optimization problems, there is also a serious limitation, especially in the solution of product combination optimization. Therefore, this paper uses an improved genetic algorithm based on greedy strategy to overcome some weaknesses of the simple genetic algorithm to achieve the optimal solution.
It is found that the study mainly focuses on the relationship between inventory and product group selection, the relationship between the stock replacement and the consumer demand, the optimal algorithm or the approximate optimal algorithm for the selection of the product group, and no scholars incorporate the consumer's information on the product to the product group selection problem. Therefore, in the era of network economy, I think it is meaningful to combine the two and make an attempt to make an attempt. The innovation of this article is mainly embodied in the following: 1., on the basis of the discrete choice model, the new elements of the net assessment score and the standard variance are added to the network environment, which is suitable for the new back of the network environment. In view, a specific consumer demand assessment model, that is, an improved MNL model, has been developed to broaden the scope of the theoretical application of the MNL model. 2. the research issues of the product classification plan are innovatively taken into consideration in the popular e-business environment, and the effect of consumer online reviews on the selection strategy of ecommerce product group is sought. Under the guidance of the tutor, by using the improved MNL discrete selection model, the customer utility expression includes the product price, the net assessment score and the standard variance, and the legacy algorithm based on the greedy strategy transforms the retailer product group selection problem into the biological genetic language. MATLAB programming is carried out, and the results of different parameters are analyzed through the construction of the model. The following conclusions are drawn:
The net evaluation score of 1. products has a positive and positive impact on the selection of the product group. The more customers care about the product online reviews, the more they are willing to choose the products with high score in the online review. In this way, more customers will be able to use the funds in several high grade products, which leads to a greater profit, although the selection of products is less. It is suggested that retailers should gradually reduce the types of products in order to facilitate customers to buy products when customers gradually pay attention to scoring.
2. there is a negative relationship between the standard variance of score and the selection of product group, that is to say, with the increase of the sensitivity coefficient of the consumers' difference, the profit of the merchant is reduced and the number of products increases. This shows that the probability of consumer choice of goods is reduced if the customer gradually emphasizes the concentration range of the net evaluation score of the commodity. In order to get more profits, businesses will have to increase the variety of products.
3. on the implementation of the algorithm, it is proved in the example of this paper that the change of the population size in the genetic algorithm will more or less affect the change of the final merchant selection result. In the process of increasing the population, the genetic evolution algebra is less generally, but the speed of each iteration is slower, the convergence is longer, and the general can find the question in less time. The optimal solution of the problem.
In this paper, under the background that consumers are increasingly affected by the information of online reviews in the network era, the model of online reviews affecting product selection is constructed, and the impact of online reviews on product decision making of network retailers is studied. It is suggested that e-commerce enterprises attach importance to the collection and evaluation of related network evaluation information, which enriches the scope of research on PAP problems and has a certain extent. However, due to the constraints of time and ability, the paper can not fully analyze the impact of online reviews on business decision making, which can only start from a few points and have a certain one-sided nature.
Although the research on the impact of online reviews on the selection of e-commerce enterprise product group is not thorough enough, it is a useful attempt after all, hoping to provide an effective reference for further research, and have certain theoretical and practical significance.
【学位授予单位】：西南财经大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：F224;F724.6

【参考文献】